Scientific Performance Audit and Convergence Metrics
This report serves as the official scientific audit for MoeaBench v0.8.0. The objective is to validate and calibrate the numerical integrity and topological fidelity of the framework's core algorithms against established mathematical benchmarks (Ground Truth).
N=200.1000 generations.M=3 objectives for volumetric analysis.| Algorithm | IGD (Mean ± Std) | GD (Mean ± Std) | SP (Mean ± Std) | EMD (Wasserstein) | H_raw | H_ratio | H_rel | Time(s) | Stabil. |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | 0.3076 | 3.1689e-01 ± 1.4e+00 | 1.6218e-02 ± 9.9e-03 | 0.0000 | 1.3185 | 0.9906 | 99.09% | 117.98 | Gen 500 |
| NSGA2 | 0.0493 | 2.4803e-01 ± 1.5e+00 | 4.3729e-02 ± 2.3e-01 | 0.0000 | 1.3310 | 1.0000 | 100.02% | 87.18 | Gen 300 |
| NSGA3 | 0.0298 | 1.4113e-02 ± 2.0e-03 | 2.7772e-03 ± 9.2e-03 | 0.0000 | 1.2642 | 0.9498 | 99.73% | 13.09 | Gen 200 |
Visual Semantics: Filled points mark solutions close to the Ground Truth, while hollow markers highlight points that remain far from the GT surface.
Clinical certification is aggregated over all available standard_runXX files for each algorithm (status mode + pass-rate by profile). The run00 layer is visualization/debug only.
Status is diagnostic and invariant: it tells you which component dominates the front (coverage/purity/shape) across 30 runs. Profiles are merely pass/fail gates on score_global. Thus a high pass-rate in Exploratory/Industry with a BIASED_SPREAD status signals “geometria boa, falta uniformidade”, and the report calls it out explicitly instead of changing the status.
| Algorithm | Geometry Status | Median scores | EXPLORATORY pass-rate | INDUSTRY pass-rate | STANDARD pass-rate | RESEARCH pass-rate | Runs | Skipped | Variability |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/5 | 5 | STABLE |
| NSGA2 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA3 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| Algorithm | IGD (Mean ± Std) | GD (Mean ± Std) | SP (Mean ± Std) | EMD (Wasserstein) | H_raw | H_ratio | H_rel | Time(s) | Stabil. |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | 0.0394 | 1.8089e-02 ± 8.2e-04 | 8.2981e-03 ± 8.8e-03 | 0.0000 | 0.7739 | 0.5815 | 97.11% | 118.74 | Gen 100 |
| NSGA2 | 0.0312 | 1.8039e-02 ± 8.3e-04 | 1.6873e-02 ± 9.1e-03 | 0.0000 | 0.8311 | 0.6244 | 97.46% | 80.07 | Gen 100 |
| NSGA3 | 0.0398 | 1.6911e-02 ± 1.4e-04 | 7.1919e-03 ± 8.7e-03 | 0.0000 | 0.7663 | 0.5757 | 96.90% | 13.27 | Gen 100 |
Visual Semantics: Filled points mark solutions close to the Ground Truth, while hollow markers highlight points that remain far from the GT surface.
Clinical certification is aggregated over all available standard_runXX files for each algorithm (status mode + pass-rate by profile). The run00 layer is visualization/debug only.
Status is diagnostic and invariant: it tells you which component dominates the front (coverage/purity/shape) across 30 runs. Profiles are merely pass/fail gates on score_global. Thus a high pass-rate in Exploratory/Industry with a BIASED_SPREAD status signals “geometria boa, falta uniformidade”, and the report calls it out explicitly instead of changing the status.
| Algorithm | Geometry Status | Median scores | EXPLORATORY pass-rate | INDUSTRY pass-rate | STANDARD pass-rate | RESEARCH pass-rate | Runs | Skipped | Variability |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA2 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA3 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| Algorithm | IGD (Mean ± Std) | GD (Mean ± Std) | SP (Mean ± Std) | EMD (Wasserstein) | H_raw | H_ratio | H_rel | Time(s) | Stabil. |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | 0.5402 | 1.6522e+00 ± 1.1e+00 | 2.1422e-01 ± 2.8e-02 | 0.0000 | 1.2829 | 0.9639 | 96.56% | 112.82 | Gen 1000 |
| NSGA2 | 0.5989 | 2.3090e+00 ± 1.4e+01 | 1.5829e-01 ± 8.7e-01 | 0.0000 | 1.3310 | 1.0000 | 100.00% | 84.55 | Gen 400 |
| NSGA3 | 0.4198 | 5.4060e-01 ± 1.7e+00 | 1.6181e-01 ± 5.1e-01 | 0.0000 | 1.3309 | 0.9999 | 99.99% | 12.43 | Gen 600 |
Visual Semantics: Filled points mark solutions close to the Ground Truth, while hollow markers highlight points that remain far from the GT surface.
Clinical certification is aggregated over all available standard_runXX files for each algorithm (status mode + pass-rate by profile). The run00 layer is visualization/debug only.
Status is diagnostic and invariant: it tells you which component dominates the front (coverage/purity/shape) across 30 runs. Profiles are merely pass/fail gates on score_global. Thus a high pass-rate in Exploratory/Industry with a BIASED_SPREAD status signals “geometria boa, falta uniformidade”, and the report calls it out explicitly instead of changing the status.
| Algorithm | Geometry Status | Median scores | EXPLORATORY pass-rate | INDUSTRY pass-rate | STANDARD pass-rate | RESEARCH pass-rate | Runs | Skipped | Variability |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA2 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA3 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| Algorithm | IGD (Mean ± Std) | GD (Mean ± Std) | SP (Mean ± Std) | EMD (Wasserstein) | H_raw | H_ratio | H_rel | Time(s) | Stabil. |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | 0.0349 | 1.8882e-02 ± 1.3e-03 | 1.3461e-02 ± 7.7e-03 | 0.0000 | 0.7705 | 0.5789 | 97.98% | 112.80 | Gen 100 |
| NSGA2 | 0.0326 | 1.7482e-02 ± 6.3e-04 | 1.8044e-02 ± 8.7e-03 | 0.0000 | 0.8314 | 0.6247 | 97.26% | 77.83 | Gen 100 |
| NSGA3 | 0.0398 | 1.6920e-02 ± 4.3e-04 | 7.2102e-03 ± 8.7e-03 | 0.0000 | 0.7718 | 0.5799 | 97.06% | 12.55 | Gen 100 |
Visual Semantics: Filled points mark solutions close to the Ground Truth, while hollow markers highlight points that remain far from the GT surface.
Clinical certification is aggregated over all available standard_runXX files for each algorithm (status mode + pass-rate by profile). The run00 layer is visualization/debug only.
Status is diagnostic and invariant: it tells you which component dominates the front (coverage/purity/shape) across 30 runs. Profiles are merely pass/fail gates on score_global. Thus a high pass-rate in Exploratory/Industry with a BIASED_SPREAD status signals “geometria boa, falta uniformidade”, and the report calls it out explicitly instead of changing the status.
| Algorithm | Geometry Status | Median scores | EXPLORATORY pass-rate | INDUSTRY pass-rate | STANDARD pass-rate | RESEARCH pass-rate | Runs | Skipped | Variability |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA2 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA3 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| Algorithm | IGD (Mean ± Std) | GD (Mean ± Std) | SP (Mean ± Std) | EMD (Wasserstein) | H_raw | H_ratio | H_rel | Time(s) | Stabil. |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | 0.0068 | 6.5874e-04 ± 3.8e-04 | 5.2277e-03 ± 8.3e-03 | 0.0000 | 0.4206 | 0.3160 | 98.91% | 111.16 | Gen 100 |
| NSGA2 | 0.0019 | 7.9038e-04 ± 1.9e-04 | 1.7806e-03 ± 1.3e-03 | 0.0000 | 0.2691 | 0.2022 | 99.65% | 82.41 | Gen 300 |
| NSGA3 | 0.0038 | 4.9039e-04 ± 3.3e-05 | 2.6192e-03 ± 2.2e-03 | 0.0000 | 0.2692 | 0.2023 | 99.16% | 11.83 | Gen 100 |
Visual Semantics: Filled points mark solutions close to the Ground Truth, while hollow markers highlight points that remain far from the GT surface.
Clinical certification is aggregated over all available standard_runXX files for each algorithm (status mode + pass-rate by profile). The run00 layer is visualization/debug only.
Status is diagnostic and invariant: it tells you which component dominates the front (coverage/purity/shape) across 30 runs. Profiles are merely pass/fail gates on score_global. Thus a high pass-rate in Exploratory/Industry with a BIASED_SPREAD status signals “geometria boa, falta uniformidade”, and the report calls it out explicitly instead of changing the status.
| Algorithm | Geometry Status | Median scores | EXPLORATORY pass-rate | INDUSTRY pass-rate | STANDARD pass-rate | RESEARCH pass-rate | Runs | Skipped | Variability |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA2 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA3 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| Algorithm | IGD (Mean ± Std) | GD (Mean ± Std) | SP (Mean ± Std) | EMD (Wasserstein) | H_raw | H_ratio | H_rel | Time(s) | Stabil. |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | 0.0560 | 8.9077e-02 ± 2.6e-02 | 1.5905e-02 ± 1.2e-02 | 0.0000 | 0.6251 | 0.4696 | 93.81% | 112.39 | Gen 200 |
| NSGA2 | 0.1586 | 2.0389e-01 ± 5.9e-01 | 1.0132e-02 ± 1.7e-02 | 0.0000 | 1.2665 | 0.9516 | 99.07% | 76.78 | Gen 400 |
| NSGA3 | 0.1930 | 2.4915e-01 ± 6.0e-01 | 1.6877e-02 ± 1.8e-02 | 0.0000 | 1.2459 | 0.9360 | 98.39% | 14.64 | Gen 600 |
Visual Semantics: Filled points mark solutions close to the Ground Truth, while hollow markers highlight points that remain far from the GT surface.
Clinical certification is aggregated over all available standard_runXX files for each algorithm (status mode + pass-rate by profile). The run00 layer is visualization/debug only.
Status is diagnostic and invariant: it tells you which component dominates the front (coverage/purity/shape) across 30 runs. Profiles are merely pass/fail gates on score_global. Thus a high pass-rate in Exploratory/Industry with a BIASED_SPREAD status signals “geometria boa, falta uniformidade”, and the report calls it out explicitly instead of changing the status.
| Algorithm | Geometry Status | Median scores | EXPLORATORY pass-rate | INDUSTRY pass-rate | STANDARD pass-rate | RESEARCH pass-rate | Runs | Skipped | Variability |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA2 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA3 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| Algorithm | IGD (Mean ± Std) | GD (Mean ± Std) | SP (Mean ± Std) | EMD (Wasserstein) | H_raw | H_ratio | H_rel | Time(s) | Stabil. |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | 0.0928 | 2.2038e-02 ± 1.7e-03 | 0.0000e+00 ± 0.0e+00 | 0.0000 | 0.0000 | 0.0000 | 0.00% | 10.00 | Gen 200 |
| NSGA2 | 0.0340 | 2.0875e-02 ± 3.9e-03 | 3.4321e-02 ± 4.3e-03 | 0.0000 | 0.6885 | 0.5173 | 96.53% | 76.05 | Gen 200 |
| NSGA3 | 0.1128 | 2.2767e-02 ± 6.6e-03 | 0.0000e+00 ± 0.0e+00 | 0.0000 | 0.0000 | 0.0000 | 0.00% | 10.00 | Gen 300 |
Visual Semantics: Filled points mark solutions close to the Ground Truth, while hollow markers highlight points that remain far from the GT surface.
Clinical certification is aggregated over all available standard_runXX files for each algorithm (status mode + pass-rate by profile). The run00 layer is visualization/debug only.
Status is diagnostic and invariant: it tells you which component dominates the front (coverage/purity/shape) across 30 runs. Profiles are merely pass/fail gates on score_global. Thus a high pass-rate in Exploratory/Industry with a BIASED_SPREAD status signals “geometria boa, falta uniformidade”, and the report calls it out explicitly instead of changing the status.
| Algorithm | Geometry Status | Median scores | EXPLORATORY pass-rate | INDUSTRY pass-rate | STANDARD pass-rate | RESEARCH pass-rate | Runs | Skipped | Variability |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA2 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/5 | 5 | STABLE |
| NSGA3 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| Algorithm | IGD (Mean ± Std) | GD (Mean ± Std) | SP (Mean ± Std) | EMD (Wasserstein) | H_raw | H_ratio | H_rel | Time(s) | Stabil. |
|---|---|---|---|---|---|---|---|---|---|
| NSGA2 | 840.6049 | 8.4035e+02 ± 1.0e+03 | 1.1544e-03 ± 2.2e-03 | 0.0000 | 0.7990 | 0.6003 | 60.03% | 30.89 | Gen 500 |
| NSGA3 | 0.0490 | 1.6853e-02 ± 9.9e-04 | 1.7490e-02 ± 1.7e-03 | 0.0000 | 0.9724 | 0.7306 | 94.99% | 15.26 | Gen 500 |
Visual Semantics: Filled points mark solutions close to the Ground Truth, while hollow markers highlight points that remain far from the GT surface.
Clinical certification is aggregated over all available standard_runXX files for each algorithm (status mode + pass-rate by profile). The run00 layer is visualization/debug only.
Status is diagnostic and invariant: it tells you which component dominates the front (coverage/purity/shape) across 30 runs. Profiles are merely pass/fail gates on score_global. Thus a high pass-rate in Exploratory/Industry with a BIASED_SPREAD status signals “geometria boa, falta uniformidade”, and the report calls it out explicitly instead of changing the status.
| Algorithm | Geometry Status | Median scores | EXPLORATORY pass-rate | INDUSTRY pass-rate | STANDARD pass-rate | RESEARCH pass-rate | Runs | Skipped | Variability |
|---|---|---|---|---|---|---|---|---|---|
| NSGA2 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/5 | 5 | STABLE |
| NSGA3 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/5 | 5 | STABLE |
| Algorithm | IGD (Mean ± Std) | GD (Mean ± Std) | SP (Mean ± Std) | EMD (Wasserstein) | H_raw | H_ratio | H_rel | Time(s) | Stabil. |
|---|---|---|---|---|---|---|---|---|---|
| NSGA2 | 1730.8555 | 1.7308e+03 ± 0.0e+00 | 0.0000e+00 ± 0.0e+00 | 0.0000 | 0.0010 | 0.0008 | 0.08% | 12.75 | Gen 100 |
| NSGA3 | 0.2067 | 6.6958e-02 ± 6.7e-03 | 8.3061e-03 ± 1.0e-03 | 0.0000 | 0.1461 | 0.1098 | 54.21% | 15.08 | Gen 500 |
Visual Semantics: Filled points mark solutions close to the Ground Truth, while hollow markers highlight points that remain far from the GT surface.
Clinical certification is aggregated over all available standard_runXX files for each algorithm (status mode + pass-rate by profile). The run00 layer is visualization/debug only.
Status is diagnostic and invariant: it tells you which component dominates the front (coverage/purity/shape) across 30 runs. Profiles are merely pass/fail gates on score_global. Thus a high pass-rate in Exploratory/Industry with a BIASED_SPREAD status signals “geometria boa, falta uniformidade”, and the report calls it out explicitly instead of changing the status.
| Algorithm | Geometry Status | Median scores | EXPLORATORY pass-rate | INDUSTRY pass-rate | STANDARD pass-rate | RESEARCH pass-rate | Runs | Skipped | Variability |
|---|---|---|---|---|---|---|---|---|---|
| NSGA2 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/5 | 5 | STABLE |
| NSGA3 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/5 | 5 | STABLE |
| Algorithm | IGD (Mean ± Std) | GD (Mean ± Std) | SP (Mean ± Std) | EMD (Wasserstein) | H_raw | H_ratio | H_rel | Time(s) | Stabil. |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | 0.3479 | 3.1650e-01 ± 9.8e-01 | 5.1073e-03 ± 2.4e-03 | 0.0000 | 1.2175 | 0.9147 | 94.40% | 117.58 | Gen 500 |
| NSGA2 | 0.0051 | 2.3259e-02 ± 1.4e-01 | 1.3109e-03 ± 3.0e-03 | 0.0000 | 1.1720 | 0.8805 | 99.83% | 82.37 | Gen 400 |
| NSGA3 | 0.0937 | 9.7682e-02 ± 4.5e-01 | 1.6700e-03 ± 1.3e-03 | 0.0000 | 1.2042 | 0.9048 | 97.41% | 12.23 | Gen 400 |
Visual Semantics: Filled points mark solutions close to the Ground Truth, while hollow markers highlight points that remain far from the GT surface.
Clinical certification is aggregated over all available standard_runXX files for each algorithm (status mode + pass-rate by profile). The run00 layer is visualization/debug only.
Status is diagnostic and invariant: it tells you which component dominates the front (coverage/purity/shape) across 30 runs. Profiles are merely pass/fail gates on score_global. Thus a high pass-rate in Exploratory/Industry with a BIASED_SPREAD status signals “geometria boa, falta uniformidade”, and the report calls it out explicitly instead of changing the status.
| Algorithm | Geometry Status | Median scores | EXPLORATORY pass-rate | INDUSTRY pass-rate | STANDARD pass-rate | RESEARCH pass-rate | Runs | Skipped | Variability |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/5 | 5 | STABLE |
| NSGA2 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA3 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| Algorithm | IGD (Mean ± Std) | GD (Mean ± Std) | SP (Mean ± Std) | EMD (Wasserstein) | H_raw | H_ratio | H_rel | Time(s) | Stabil. |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | 0.9019 | 4.0960e-05 ± 4.5e-07 | 3.5453e-03 ± 8.4e-06 | 0.0000 | 0.1921 | 0.1443 | 49.90% | 115.59 | Gen 100 |
| NSGA2 | 0.0017 | 1.2828e-04 ± 2.0e-05 | 1.0080e-02 ± 8.1e-03 | 0.0000 | 0.3841 | 0.2886 | 99.72% | 72.52 | Gen 100 |
| NSGA3 | 0.0031 | 1.2220e-04 ± 1.5e-05 | 1.4596e-02 ± 1.2e-02 | 0.0000 | 0.3844 | 0.2888 | 99.54% | 13.38 | Gen 200 |
Visual Semantics: Filled points mark solutions close to the Ground Truth, while hollow markers highlight points that remain far from the GT surface.
Clinical certification is aggregated over all available standard_runXX files for each algorithm (status mode + pass-rate by profile). The run00 layer is visualization/debug only.
Status is diagnostic and invariant: it tells you which component dominates the front (coverage/purity/shape) across 30 runs. Profiles are merely pass/fail gates on score_global. Thus a high pass-rate in Exploratory/Industry with a BIASED_SPREAD status signals “geometria boa, falta uniformidade”, and the report calls it out explicitly instead of changing the status.
| Algorithm | Geometry Status | Median scores | EXPLORATORY pass-rate | INDUSTRY pass-rate | STANDARD pass-rate | RESEARCH pass-rate | Runs | Skipped | Variability |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/5 | 5 | STABLE |
| NSGA2 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA3 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| Algorithm | IGD (Mean ± Std) | GD (Mean ± Std) | SP (Mean ± Std) | EMD (Wasserstein) | H_raw | H_ratio | H_rel | Time(s) | Stabil. |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | 0.0238 | 2.7892e-03 ± 8.8e-04 | 4.2399e-03 ± 9.4e-03 | 0.0000 | 0.7761 | 0.5831 | 97.65% | 115.10 | Gen - |
| NSGA2 | 0.0016 | 3.0564e-03 ± 1.2e-04 | 2.0067e-03 ± 1.4e-03 | 0.0000 | 0.7945 | 0.5969 | 100.22% | 75.93 | Gen 800 |
| NSGA3 | 0.0042 | 3.1938e-03 ± 1.6e-04 | 3.0014e-03 ± 2.3e-03 | 0.0000 | 0.7982 | 0.5997 | 99.99% | 12.23 | Gen 100 |
Visual Semantics: Filled points mark solutions close to the Ground Truth, while hollow markers highlight points that remain far from the GT surface.
Clinical certification is aggregated over all available standard_runXX files for each algorithm (status mode + pass-rate by profile). The run00 layer is visualization/debug only.
Status is diagnostic and invariant: it tells you which component dominates the front (coverage/purity/shape) across 30 runs. Profiles are merely pass/fail gates on score_global. Thus a high pass-rate in Exploratory/Industry with a BIASED_SPREAD status signals “geometria boa, falta uniformidade”, and the report calls it out explicitly instead of changing the status.
| Algorithm | Geometry Status | Median scores | EXPLORATORY pass-rate | INDUSTRY pass-rate | STANDARD pass-rate | RESEARCH pass-rate | Runs | Skipped | Variability |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA2 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA3 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| Algorithm | IGD (Mean ± Std) | GD (Mean ± Std) | SP (Mean ± Std) | EMD (Wasserstein) | H_raw | H_ratio | H_rel | Time(s) | Stabil. |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | 0.5025 | 1.4101e-01 ± 7.8e-01 | 8.9064e-04 ± 1.6e-03 | 0.0000 | 0.6949 | 0.5221 | 53.27% | 121.37 | Gen 300 |
| NSGA2 | 0.0044 | 2.4544e-02 ± 1.5e-01 | 3.3861e-03 ± 3.4e-03 | 0.0000 | 1.1632 | 0.8739 | 99.64% | 82.92 | Gen 200 |
| NSGA3 | 0.0233 | 1.8991e-02 ± 1.2e-01 | 1.3051e-02 ± 4.9e-02 | 0.0000 | 1.3061 | 0.9813 | 99.68% | 12.23 | Gen 200 |
Visual Semantics: Filled points mark solutions close to the Ground Truth, while hollow markers highlight points that remain far from the GT surface.
Clinical certification is aggregated over all available standard_runXX files for each algorithm (status mode + pass-rate by profile). The run00 layer is visualization/debug only.
Status is diagnostic and invariant: it tells you which component dominates the front (coverage/purity/shape) across 30 runs. Profiles are merely pass/fail gates on score_global. Thus a high pass-rate in Exploratory/Industry with a BIASED_SPREAD status signals “geometria boa, falta uniformidade”, and the report calls it out explicitly instead of changing the status.
| Algorithm | Geometry Status | Median scores | EXPLORATORY pass-rate | INDUSTRY pass-rate | STANDARD pass-rate | RESEARCH pass-rate | Runs | Skipped | Variability |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/5 | 5 | STABLE |
| NSGA2 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA3 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| Algorithm | IGD (Mean ± Std) | GD (Mean ± Std) | SP (Mean ± Std) | EMD (Wasserstein) | H_raw | H_ratio | H_rel | Time(s) | Stabil. |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | 0.0304 | 6.7792e-03 ± 1.4e-04 | 1.6212e-02 ± 2.0e-02 | 0.0000 | 0.7066 | 0.5309 | 96.80% | 113.39 | Gen 100 |
| NSGA2 | 0.0142 | 9.1868e-03 ± 4.9e-03 | 2.0535e-02 ± 8.9e-03 | 0.0000 | 1.1082 | 0.8326 | 99.13% | 81.79 | Gen 100 |
| NSGA3 | 0.0239 | 6.2888e-03 ± 1.8e-04 | 1.8744e-02 ± 1.0e-02 | 0.0000 | 0.7885 | 0.5924 | 97.87% | 12.51 | Gen 200 |
Visual Semantics: Filled points mark solutions close to the Ground Truth, while hollow markers highlight points that remain far from the GT surface.
Clinical certification is aggregated over all available standard_runXX files for each algorithm (status mode + pass-rate by profile). The run00 layer is visualization/debug only.
Status is diagnostic and invariant: it tells you which component dominates the front (coverage/purity/shape) across 30 runs. Profiles are merely pass/fail gates on score_global. Thus a high pass-rate in Exploratory/Industry with a BIASED_SPREAD status signals “geometria boa, falta uniformidade”, and the report calls it out explicitly instead of changing the status.
| Algorithm | Geometry Status | Median scores | EXPLORATORY pass-rate | INDUSTRY pass-rate | STANDARD pass-rate | RESEARCH pass-rate | Runs | Skipped | Variability |
|---|---|---|---|---|---|---|---|---|---|
| MOEAD | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA2 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |
| NSGA3 | UNDEFINED | n/a | 0.0% | 0.0% | 0.0% | 0.0% | 0/30 | 30 | STABLE |